Scalable Context-aware Services

Most of the prior work in ubiquitous computing focused on small Context-Aware Services (CASs). However, there is a trend to build also large-scale applications where many context sources, CASs, and users are spread over a large area. These kinds of CASs are subject to strong scalability problems. In my work I show how scalable CASs can be built. The main idea is to dynamically replicate and distribute CASs within the main scalability infrastructure of the world wide web, namely Content Delivery Networks (CDNs). Unfortunately, today’s CDNs are not able to deliver CASs mainly due to two reasons: First, it is not solved how replicas of CASs can access context information from within the CDN. Second, the distribution algorithms employed in CDNs are not well suited for distributing CASs. In my work I address these two issues. I designed an integration layer that resides on the servers of the CDNs, binds to the heterogeneous regional context sources, and provides context information in a unified form to the replicas of the CASs. Furthermore, I developed a new heuristics for the distribution of CASs within CDNs that bases its replica placement decisions on a profit maximization calculation and uses statistical process control techniques.